diff --git a/build.py b/build.py index 57acf198f6..8b47cd78f0 100755 --- a/build.py +++ b/build.py @@ -74,11 +74,11 @@ "release_version": "2.67.0dev", "triton_container_version": "26.03dev", "upstream_container_version": "26.02", - "ort_version": "1.24.1", - "ort_openvino_version": "2025.4.1", - "standalone_openvino_version": "2025.4.1", + "ort_version": "1.24.2", + "ort_openvino_version": "2026.0.0", + "standalone_openvino_version": "2026.0.0", "dcgm_version": "4.5.2-1", - "vllm_version": "0.15.1", + "vllm_version": "0.16.0", "rhel_py_version": "3.12.3", } diff --git a/qa/common/gen_qa_model_repository b/qa/common/gen_qa_model_repository index f7cd2b0cbe..bcd00c90d8 100755 --- a/qa/common/gen_qa_model_repository +++ b/qa/common/gen_qa_model_repository @@ -344,7 +344,18 @@ chmod -R 777 $TRITON_MDLS_QA_TRT_DATA_DEPENDENT_MODEL if [ -d "/usr/src/tensorrt/samples/python/onnx_custom_plugin" ]; then cd /usr/src/tensorrt/samples/python/onnx_custom_plugin else - git clone -b release/\$( echo \$TRT_VERSION | cut -d . -f -2) --depth 1 https://github.com/NVIDIA/TensorRT.git /workspace/TensorRT + TRT_BRANCH=\$(echo \$TRT_VERSION | cut -d . -f -2) + if ! git clone -b release/\${TRT_BRANCH} --depth 1 https://github.com/NVIDIA/TensorRT.git /workspace/TensorRT; then + MAJOR=\$(echo "\$TRT_BRANCH" | cut -d . -f 1) + MINOR=\$(echo "\$TRT_BRANCH" | cut -d . -f 2) + if [ -n "\$MINOR" ] && [ "\$MINOR" -gt 0 ] 2>/dev/null; then + TRT_BRANCH="\${MAJOR}.\$((MINOR - 1))" + echo "Fallback: cloning TensorRT release/\${TRT_BRANCH} (previous minor)" + git clone -b release/\${TRT_BRANCH} --depth 1 https://github.com/NVIDIA/TensorRT.git /workspace/TensorRT + else + exit 1 + fi + fi cd /workspace/TensorRT/samples/python/onnx_custom_plugin fi rm -rf build && mkdir build && \ @@ -441,7 +452,7 @@ if [ "$TRITON_MODELS_USE_DOCKER" -eq 1 ] && which docker ; then if [[ "aarch64" != $(uname -m) ]] ; then - log_message.status "docker run: $OPENVINOSCRIPT" + log_message.info "docker run: $OPENVINOSCRIPT" docker run \ --rm \ --label RUNNER_ID=$RUNNER_ID \ @@ -457,7 +468,7 @@ if [ "$TRITON_MODELS_USE_DOCKER" -eq 1 ] && which docker ; then fi fi # [[ "aarch64" != $(uname -m) ]] - log_message.status "docker run: $ONNXSCRIPT" + log_message.info "docker run: $ONNXSCRIPT" docker run \ --rm \ --label RUNNER_ID=$RUNNER_ID \ @@ -475,7 +486,7 @@ if [ "$TRITON_MODELS_USE_DOCKER" -eq 1 ] && which docker ; then log_message.status "docker pull: $PYTORCH_IMAGE" docker pull $PYTORCH_IMAGE - log_message.status "docker run: $TORCHSCRIPT" + log_message.info "docker run: $TORCHSCRIPT" docker run \ --rm \ --label RUNNER_ID=$RUNNER_ID \ @@ -494,7 +505,7 @@ if [ "$TRITON_MODELS_USE_DOCKER" -eq 1 ] && which docker ; then log_message.status "docker pull: $TENSORRT_IMAGE" docker pull $TENSORRT_IMAGE - log_message.status "docker run: $TRTSCRIPT" + log_message.info "docker run: $TRTSCRIPT" docker run \ --rm \ --label RUNNER_ID=$RUNNER_ID \ @@ -594,7 +605,7 @@ elif [ "$TRITON_MODELS_USE_ENROOT" -eq 1 ] && which enroot ; then log_message.status "enroot create: pytorch.$CI_JOB_ID" enroot create --name pytorch.$CI_JOB_ID /tmp/pytorch.$CI_JOB_ID.enroot.sqsh - log_message.status "enroot start: pytorch.$CI_JOB_ID" + log_message.info "enroot start: pytorch.$CI_JOB_ID" enroot start --rw -m /tmp:/tmp pytorch.$CI_JOB_ID bash -xe $TRITON_MDLS_SRC_DIR/$TORCHSCRIPT if [ $? -ne 0 ]; then log_message.error "enroot start: ${TORCHSCRIPT} failed" @@ -605,7 +616,7 @@ elif [ "$TRITON_MODELS_USE_ENROOT" -eq 1 ] && which enroot ; then enroot import --output /tmp/tensorrt.$CI_JOB_ID.enroot.sqsh docker://$TENSORRT_IMAGE log_message.status "enroot create: tensorrt.$CI_JOB_ID" enroot create --name tensorrt.$CI_JOB_ID /tmp/tensorrt.$CI_JOB_ID.enroot.sqsh - log_message.status "enroot start: tensorrt.$CI_JOB_ID" + log_message.info "enroot start: tensorrt.$CI_JOB_ID" enroot start --rw -m /tmp:/tmp tensorrt.$CI_JOB_ID bash -xe $TRITON_MDLS_SRC_DIR/$TRTSCRIPT if [ $? -ne 0 ]; then log_message.error "enroot start: ${TRTSCRIPT} failed"